What problem does it solve? Writing design specs without confirmed requirements or systematic review leads to rework, contradictions, and documents that cannot be turned into implementation plans. This Skill enforces a strict, auditable workflow so every design spec is grounded in confirmed requirements and passes multi-dimensional review before use. ## Core Features & Use Cases - Mandatory pre-work: Reads project docs.md rules, the app feature list, and recent existing specs before any writing begins, ensuring naming, structure, and style consistency. - Grilling before writing: Conducts one-question-at-a-time requirement interrogation covering goals, scope, data models, compatibility, and acceptance criteria before drafting. - Three-agent parallel review: Dispatches three sub-agents simultaneously to check completeness and docs.md compliance, planability and YAGNI scope, and verifiability with UI observability. - Per-step git commits: Every step produces a committed artifact, creating an auditable history with Conventional Commits messages. - Use Case: When starting a new feature or API migration, invoke this Skill to grill requirements, generate a 12-section design spec via a sub-agent, run three parallel reviews, and confirm with the user one question at a time before moving to visual prototypes or test case tables. ## Quick Start Ask the AI to write a design spec for the new feature using the dd-writing-design-specs workflow, starting by reading the project docs rules and grilling the requirements one question at a time.
